Abstract #0312

Physics-informed deep neural network for tri-exponential intravoxel incoherent motion fitting in non-alcoholic fatty liver disease.

Marian A. Troelstra¹, Julia J. Witjes², Anne-Marieke van Dijk², Anne Linde Mak², Jurgen H. Runge¹, Joanne Verheij³, Max Nieuwdorp², Adriaan G. Holleboom², Aart J. Nederveen¹, and Oliver J. Gurney-Champion¹

¹Department of Radiology and Nuclear Medicine, Amsterdam UMC, location AMC, Amsterdam, Netherlands, ²Department of Internal and Vascular Medicine, Amsterdam UMC, location AMC, Amsterdam, Netherlands, ³Department of Pathology, Amsterdam UMC, location AMC, Amsterdam, Netherlands

In this study we have developed an unsupervised physics-informed deep neural network (IVIM₃-NET) to fit a tri-exponential model to intravoxel incoherent motion (IVIM) data from 35 non-alcoholic fatty liver disease (NAFLD) patients. Diagnostic performance was compared to a tri-exponential least squares (LSQ) fit. Visually, IVIM₃-NET showed high-quality parameter maps with less noise than the LSQ-fit. IVIM₃-NET showed slightly higher correlations between fit parameters and histology and more significant differences between levels of fibrosis and inflammation than the LSQ-fit. Correlations between f₂ and fibrosis and inflammation grade, potentially highlighting NAFLD-induced vascular changes, warrant further investigation of the IVIM₃-NET in NAFLD patients.

This abstract and the presentation materials are available to members only; a login is required.

Join Here